Modeling Abrupt Climate Change as the Interaction Between Sea Ice Extent and Mean Ocean Temperature Under Orbital Insolation Forcing - Abrupt Climate Change: Mechanisms, Patterns, and Impacts - page 71

Figure A2. (a) Time series of Arctic and Antarctic surface temperatures obtained with ECBILT-CLIO compared to (b) the
actual ice core records (NGRIP and EDC3). Notice that though the simulation is in a different time scale (self-similarity in
time is a property of the ice core time series as described by Rial and Yang [2007]) than the whole data window (simulation
of a 100 kyr long time series takes up to 14 months of continuous computing), the phase relationships and waveforms are
fully consistent with those exhibited by the data. In both Figures A2a and A2b, the time integral of the northern record
reproduces the southern one, and conversely, the Hilbert transform of the southern record reproduces the northern one.
These mathematical operations are just convenient tools; the actual relationship is a π /2 shift with which Antarctica ' s
temperature leads Greenland [Steig, 2006]. Nevertheless, the implications are important because they reveal that there is a
fundamental connection between the climates of the polar regions, akin to synchronization (J. A. Rial, Synchronization of
climatic change between the polar regions as a feasible origin of the bipolar seesaw, submitted to Climate Change, 2011).
Note that the ECBILT-CLIO simulation is not forced with the insolation, which would increase the ~1.5 kyr fluctuations to
5 kyr or more, as can be seen in Figures 3, 5, 6, and 7.

included in ECBILT-CLIO, the D-O-like oscillations fade.
But since ECBILT-CLIO does not grow or melt continental
ice and its ice sheets are fixed boundary conditions, we opt to
describe them as flat, high-albedo surface areas because our
objective is to study the high-frequency, D-O-like fluctua-
tions that the code produces in this case and how other
variables in the code react to these fluctuations. Similarly, it
is well known that the Langevin model of Figure 3 needs to
be tuned in order to produce bistable fluctuations. Moreover,
these are the result of integrating Newton ' s equations repre-
senting random jumps of a small ball between two “ poten-
tial ” wells, yet that hardly prevents its use or that of similar
equations [e.g., Sakai and Peltier, 1997] as useful simulators
of climate variability. Lorenz [1976] has clearly illustrated
this. Thus, if the D-O-like oscillation in ECBILT-CLIO is an

artifact, as far as our experiments are concerned, it must be a
benign one.
An extensive literature provides a good source for the
scope and limitations of ECBILT-CLIO. A necessarily in-
complete list includes Widmann et al. [2009], Timmermann
et al. [2009], Weber et al. [2007], Yanase and Abe-Ouchi
[2007], Schulz et al. [2007], Fluckiger et al. [2006], and
Knutti et al. [2004].

Acknowledgments. Steven Meyers, Austin Milt, Bruce Bills, and
two anonymous reviewers made the original manuscript intelligible.
This research was sponsored by a grant from the National Science
Foundation Paleoclimate Program. Supplementary funding was
provided by a grant from the National Geographic committee for
research and exploration.